Tracer Advection is substantially different for GEOS

[FlorianDeconinck]

GEOS differs from FV3GFS.

A quick glance at code diff yields:

• PyFV3 uses cmax = 2, remove this workaround
• dpa optional calculation
• nsplt/ksplt are not computed the same way
• Added offline_tracer_advection case, triggered by AdvCore_Advection. Out of scope for this (we are targeting online dycore) but we should get to the bottom of it in term of usage.

Strategy should be to diff the original port (https://github.com/ai2cm/fv3gfs-fortran) with the current code to get a sense. Then use serialbox to generate or re-generate the translate tests.

Note: feel free to break this further into G1.8.X

Parent: GEOS-ESM/SMT-Nebulae#55

---

• TracerAdvection in PyFV3 validates under GEOS translate tests (without breaking the original test case)

[FlorianDeconinck]

Breakdown of the tasks:

• Add options to advection that leave mass and courant untouched
• Write cmax calculation
• Make n_split a column field & move n_split loop within the tracers stencils
• Mass fixes

[FlorianDeconinck]

Porting work done and verifying.

Still broken:

• Orchestration